-
Notifications
You must be signed in to change notification settings - Fork 623
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
disable_tracing_hostname for URL exclude lists not covering reasonable use cases #779
Comments
I think originally, the logic taken from OpenCensus was to cover the use cases that were most common (users would match exact urls that they do not want to be traced with hostnames, or urls under a certain path would not want to be traced). I'm not to certain whether or not these included things like param. Regardless, if the new goal with OpenTelemetry exclude listing is to provide a mechanism in which to cover ALL uses cases, and not be very tedious for users to use (theoretically we can do exact matching for everything), then yes, the current design is lacking in some areas. However, I don't know if matching all use cases makes sense practicality wise (how often would someone want to exclude Could this be as simple as removing Also, what did you mean by " ie for endpoint /hello/{helloid}/info a user could just add /hello/{helloid}/info to the exclude list instead of needing to write a regex."? |
I think removing If the user isn't supposed to use regex then they can't generalize paths to be endpoints, for example if they wanted to block |
@ocelotl Thoughts? |
Well, after taking into consideration all the details like params and protocols, I am more inclined to use a single environment variable that uses a regex and let the user just write the regex to do the exclusions. It is the most powerful approach, since it should cover all possible cases. While regexes can be not too user friendly, I guess that in most of the cases they won't be too different from what we have now, for example, the regex to exclude |
Sounds good. I think we'll use this as the desired solution. |
* chore: fix API docs * update WEB Readme: WebTracer => WebTracerProvider
There are two URL exclude lists for webservers:
disable_tracing_path
Disables paths based on a regex after (not including) the hostname. Should be able to include/exclude any combination of paths, but does not include hostname
disable_tracing_hostname
Disable path based on the exact URL being included in a list.
disable_tracing_hostname
seems to be significantly less useful thandisable_tracing_path
, since if there is any URL parameters in the endpoint, eghttp://site.com/abc?param=2
it will not have a use (specific parameter values can be disabled by disable_tracing_path, if desired). Personally I would supportdisable_tracing_hostname
performing a regex on everything before the path, or if there are cases where we need a combination of hostname + path, it could regex the whole URL.For example, if a user wanted to exclude all endpoints that were not called with
http://
orhttps://
, this could not be done with the current exclude lists. Or if the user wanted to exclude requests toadmin.website.com/path
but notwebsite.com/path
, this could also not be done.There may also be value to using the endpoint pattern instead of the actual URL, ie for endpoint
/hello/{helloid}/info
a user could just add/hello/{helloid}/info
to the exclude list instead of needing to write a regex. However this would also givedisable_tracing_path
less functionality with regards to excluding specific values passed in, so this would need some more thinking.Thoughts?
The text was updated successfully, but these errors were encountered: